Introduction
Covalent organic frameworks (COFs) are a class of crystalline, porous, and two-dimensional
(2D) or three-dimensional organic polymer materials. They are composed of lightweight
elements (C, H, O, N, B) and synthesized via dynamic covalent bond formation.[1 ]
[2 ]
[3 ] Different types of linkages such as imine,[4 ] hydrazone,[5 ] azine,[6 ] boronate ester,[1 ] and triazines[7 ] can be found in these materials. After seminal work by Yaghi and co-workers in 2005,[1 ] numerous COFs have been developed. These materials have been employed for various
applications including catalysis,[8 ] gas storage,[9 ] gas separations,[10 ] energy storage,[11 ] and sensing,[12 ] due to their promising properties such as well-defined structures,[3 ]
[13 ] high surface area,[9 ] tunable pore size,[14 ] and low density.[9 ]
The general mechanisms of COF polymerization and assembly have received considerable
attention over the past few years[15 ]
[16 ] in order to help researchers understand why certain combinations of monomers form
crystalline COFs over others. Hydrazone-based COFs have been studied extensively due
to their physicochemical stability compared to imine, or boronate ester-linked COFs.[5 ]
[17 ]
[18 ] 2,5-Diethoxyterephthalohydrazide was used as the hydrazide linker in the first report
of hydrazone-linked COFs by Yaghi and coworkers.[5 ] Since then many other hydrazone COFs have been reported using this strategy.[17 ]
[19 ]
[20 ]
[21 ] In almost all these reports, hydrazone COFs were synthesized with monomers containing
alkoxy (methoxy, ethoxy, propoxy) or allyloxy side chains.[17 ]
[20 ]
[21 ]
[22 ] However, the role of these side chains in the initial polymerization of hydrazone
COFs has not been fully elucidated. It is clear that these side chains play a more
crucial role in the formation of crystalline materials than in COFs with other types
of linkages such as imines, β-ketoenamines, or boronate esters. Attempts to synthesize
hydrazone COFs from unsubstituted or alkyl chain substituted hydrazides have so far
been unsuccessful and typically result in non-porous or amorphous polymers (or both).[19 ]
[20 ]
[22 ] Previous studies have reported the importance of these side chains through mechanistic
and spectroscopic studies and found that these side chains can facilitate the formation
of intramolecular hydrogen bonding, which causes the restriction of intramolecular
bond rotation.[20 ]
[22 ]
[23 ]
As shown in [Figure 1a ] and [b ], the presence of alkoxy side chains enables improved intra- and interlayer hydrogen
bonding. In contrast, monomers without any substituent can experience free intramolecular
bond rotation resulting in more flexibility in the sheet structure that can potentially
reduce crystallinity. If these side chains serve to rigidify the 2D layers or enhance
the strength of the interactions between the sheets, then it is possible that the
lack of these side chains could also reduce the stability of the COFs to activation.
Indeed, previous studies have shown that the method of activation used on a COF plays
a pivotal role in its bulk porosity and crystallinity. The most widely used activation
method reported in the literature so far is solvent activation.[24 ] Recently, more mild activation methods have been reported, which highlights the
importance of proper mild activation of COFs towards their crystallinity and porosity.
These methods include the activation with ultralow surface tension solvents, nitrogen-flow
activation, and the use of supercritical fluid such as supercritical carbon dioxide
(scCO2 ).[24 ]
[25 ]
[26 ]
Figure 1 Inter- and/or intralayer hydrogen-bonding interactions that are available for side-chain
functionalized (a) and unfunctionalized (b) hydrazone-linked COFs. (c) Illustration
of pathways by which a crystalline COF can become amorphous.
Mechanistically, the loss of long-range order of the framework during COF formation
and isolation can be associated with two possible hypotheses, random displacement
of the COF layers or pore collapse during the activation process, as illustrated in
[Figure 1c ]. A recent study revealed that the displacement of COF layers upon exposure to solvent
vapors causes weakening of the interlayer interactions.[25 ] Another report has shown that hydrazone COFs can be delaminated in polar organic
solvents with ultrasonication.[18 ] Since scCO2 has a lower surface tension compared with conventional organic solvents, the strength
of the capillary forces is much lower (and much less damaging) on the COF structure
when it is removed.[24 ]
[25 ]
[26 ]
Herein, we report the synthesis and characterization of three side-chain-free hydrazone
COFs that are both porous and crystalline upon scCO2 activation. Synthesis of side-chain-free hydrazone COFs will expand the scope of
hydrazone COF chemistry. To the best of our knowledge, this is the first report on
scCO2 activation of hydrazone COFs, which were synthesized from unsubstituted linkers.
Results and Discussion
In this study two different tritopic linkers, 5'-(4-formylphenyl)-[1,1':3',1''-terphenyl]-4,4''-dicarbaldehyde
(TFPB ) ([Scheme S1 ]) and 4,4',4''-(1,3,5-triazine-2,4,6-triyl)tribenzaldehyde (TFPT ) ([Scheme S2 ]), and one tetratopic linker, 4,4',4'',4'''-(pyrene-1,3,6,8-tetrayl)tetrabenzaldehyde
(Py ) ([Scheme S3 ]), were selected to synthesize hydrazone COFs with the linear ditopic linker terephthalohydrazide
(DHz ). The aldehyde monomers were synthesized according to previously published literature.[18 ]
[21 ]
[27 ] Three different COFs were synthesized via the solvothermal method using optimized
conditions ([Scheme 1 ]). TFPB-DHz COF and TFPT-DHz COF were synthesized in a mixture of dioxane and mesitylene (1:9 v/v) with 6 M acetic
acid as the catalyst at 120 °C for 72 h. Py-DHz COF was synthesized in a mixture of n -butanol and o -dichlorobenzene (1:9 v/v) with 6 M acetic acid as the catalyst at 120 °C for 120 h.
The activation of the COFs was carried out with both scCO2 and conventional methods. The COFs were obtained as yellow powders and were insoluble
in common organic solvents such as acetone, ethanol, THF, and DMF. We synthesized
a model compound ([Scheme S4 ]) for structural conformation of the synthesized COFs.
Scheme 1 Synthesis of TFPB-DHz COF, Py-DHz COF and TFPT-DHz COF through condensation of linear linker DHz with TFPB , Py and TFPT , respectively. Pore sizes shown were calculated from a computational model of the
structure. Images of the COF powders are shown in the insets.
The synthesized COFs were characterized with Fourier transform infrared (FT-IR) spectroscopy.
The corresponding IR spectra of COFs, monomers, and the model compound were collected
([Figures S3–S5 ]). The stretching modes observed for TFPT-DHz COF and TFPB-DHz COF at 1566–1605 and 1273 cm−1 are characteristic of νC = N moieties, which confirms the formation of C = N linkages
in both COFs. Also, in Py-DHz COF, C = N stretching vibrations were observed at 1605 and 1273 cm−1 . These stretching frequencies were comparable with the model compound's νC = N stretching
modes observed at 1605 and 1280 cm−1 . These results indicate the formation of hydrazone moieties from the polycondensation
between hydrazide and aldehyde linkers. Furthermore, the disappearance of C = O vibrations
of aldehyde monomers (at 1682, 1697, and 1689 cm−1 for TFPB , TFPT , and Py respectively) and the amino group vibration bands (at 3317 and 3209 cm−1 ) of DHz in all COFs were clear indication of the absence of starting monomers. Also, the
carbonyl (νC = O) stretching vibrations of hydrazone linkage of TFPT-DHz COF, TFPB-DHz COF, Py-DHz COF, model compound, and DHz monomer were observed at 1659, 1659, 1659, 1651, and 1612 cm−1 respectively. The change in the νC = O stretching frequency to a higher wavenumber
in COFs relative to the monomer and the model compound can be attributed to the decrease
in stability of the C = O bond due to an increase in conjugation. Moreover, the N–H
stretching frequency of hydrazone moieties for TFPB-DHz COF and TFPT-DHz COF was observed at 3201 cm−1 , while it was at 3178 cm−1 for Py-DHz COF. For TFPT-DHz COF, the triazine ring breath was observed at 810 cm−1 , which matched with the literature reported value.[21 ]
The crystallinity of all synthesized COFs was investigated by powder X-ray diffraction
(PXRD) analysis. The diffraction patterns of COFs are depicted in [Figure 2 ]. scCO2 -activated TFPB-DHz COF, TFPT-DHz COF, and Py-DHz COF exhibited excellent crystallinity, while the corresponding conventional solvent
activated COFs appeared to be amorphous ([Figures S8–S10 ]). The scCO2 -activated COFs showed multiple diffraction peaks in their diffraction profiles, which
indicates the long-range order in the structures. In the PXRD pattern of TFPB-DHz COF, an intense diffraction peak was observed at 2.5° corresponding with the (100)
plane, while additional diffraction peaks were observed at 4.3°, 5.2°, 6.6°, 8.5°,
10.0°, and 25.0°, which originated from the (200), (210), (220), (230), (310), and
(001) planes, respectively ([Figure 2a ]). Similarly, an intense peak was seen at 2.6° for the TFPT-DHz COF, which corresponds to the (100) plane. In addition to the peak at 2.6°, multiple
diffraction peaks were observed at 3.8°, 4.3°, 5.1°, 6.6°, 8.7°, 10.5°, and 26.3°.
These can be assigned to the (110), (200), (210), (220), (230), (310), and (001) facets,
respectively ([Figure 2b ]). The PXRD pattern of the Py-DHz COF showed diffraction peaks at 3.3°, 5°, 6.67°, 10°, 18.8°, 23°, and 28.4°, which
correspond to the (110), (020), (220), (330), and (001) crystal planes ([Figure 2c ]). These results indicate that scCO2 -activated COFs were highly crystalline. The interlayer distances (d spacing values) corresponding to the 001 reflections were calculated to be 3.56,
3.39, and 3.15 Å for TFPB-DHz COF, TFPT-DHz COF, and Py-DHz COF, respectively. This observation can be attributed to the close packing of the
Py-DHz COF layers compared to the TFPB-DHz COF and TFPT-DHz COF sheets. Molecular modeling and Pawley refinements of TFPB-DHz COF, TFPT-DHz COF, and Py-DHz COF were performed using BIOVA Material Studio 2019 software package. The hexagonal
lattice structural models of both TFPB-DHz COF and TFPT-DHz COF were built using the P6/m space group, with AA-eclipsed stacking model and minimized using the universal force
field (UFF). The experimental PXRD patterns of both COFs had good agreement with the
simulated patterns for the eclipsed stacking model. The (R
p , R
wp ) values converged to (1.55%, 2.25%) and (3.23%, 4.38%) for TFPB-DHz COF and TFPT-DHz COF, respectively.
Figure 2 Experimental (green), Pawley refined (red), difference plot (black), simulated eclipsed
stacking (blue) PXRD patterns of (a) TFPB-DHz COF, (b) TFPT-DHz COF, and (c) Py-DHz COF. Inset: refined AA-stacking structures. C, ash; H, white; N, blue; O, red.
The Pawley-refined unit cell parameters of the TFPB-DHz COF were a = b = 45.32 Å, c =3.65 Å, α = 90°, β = 90°, and γ = 120°, whereas for
the TFPT-DHz COF those were a = b = 41.14 Å, c = 3.51 Å, α = 90°, β = 90°, and γ = 120°. The monoclinic
lattice structural model of the Py-DHz COF was built using the C2/m space group, with AA-eclipsed stacking model and minimized using UFF. The experimental
PXRD pattern of the Py-DHz COF had good agreement with the simulated pattern for the eclipsed stacking model.
The R
p , R
wp values converged to 2.81%, 3.59%, respectively. The Pawley-refined unit cell parameters
of the Py-DHz COF were a = 36.93 Å, b = 43.75 Å, c =3.95 Å, α = γ = 90°, and β = 119.2°.
The surface area of the COFs was measured by N2 adsorption experiments at 77 K. The adsorption–desorption isotherms are shown in
[Figure 3 ]. The isotherms of both TFPT-DHz COF and TFPB-DHz COF exhibited a type-IV isotherm, indicating the mesoporous nature of these materials.
However, the Py-DHz COF possesses a type-I isotherm, characteristic of a microporous material. The Brunauer − Emmett − Teller
(BET) surface areas of TFPT-DHz COF, TFPB-DHz COF, and Py-DHz COF were found to be 1199, 790, and 932 m2 /g, respectively. The higher surface area of the TFPT-DHz COF compared with that of the TFPB-DHz COF can be attributed to the more favorable stacking interactions in TFPT-DHz COF layers due to the highly planar structure of the triazine-centered TFPT .[21 ] Interestingly, the conventionally activated COFs were amorphous in nature with the
surface area of 55 m2 /g (TFPB-DHz COF, [Figure S6 ]), 0 m2 /g (TFPT-DHz COF, [Figure S6 ]), and 256 m2 /g (Py-DHz COF, [Figure S6 ]). Hence this study reveals the importance of processing COFs under mild activation
conditions to preserve their crystallinity and porosity. Furthermore, it indicates
that the mechanism of the COF polymerization reaction is not affected by the presence
of the alkoxy side chains.
Figure 3 N2 adsorption (solid circles) and desorption (open circles) isotherms of COFs. Inset:
the pore size distributions of COFs.
The pore size distributions were determined using nonlocal density functional theory
(NLDFT). The pore size distributions for each COF are shown in [Figure 3 ]. TFPB-DHz COF and TFPT-DHz COF have pore sizes confined around 37 and 41 Å, respectively. These pore sizes are
in good agreement with the theoretical pore sizes obtained from crystal modeling of
an AA-eclipsed stacking structure. The Py-DHz COF was measured to have a pore size of 17 Å and showed a type I isotherm. However,
the theoretical pore diameter obtained from the computational model suggests the formation
of a mesoporous pore. This difference can be explained by the fact that Py-DHz COF contains non-planar tetraphenylpyrene monomer units that can experience larger
offsets between the layers than a typical COF with more planarized monomer units.[28 ]
The COF polymerization process has been studied by Marder and co-workers, and they
found that the disordered thin, crystalline COF layers were formed within the aggregates
at the very early stage of the COF polymerization.[16 ] Although this work was performed with imine COFs, these principles appear to be
relevant in the case of hydrazone COFs, as well as hydrazone COFs, when carefully
activated, retain their crystalline structure. The nature of the loss of crystallinity
upon activation is less clear, especially since we have observed it with hydrazone
COFs that are similar to those known to readily delaminate in polar solvents[18 ] (TFPT-DHz and TFPB-DHz COFs), as well as Py-DHz COF which is made with a tetraphenylpyrene monomer that is well known to form highly
correlated stacking interactions in 2D-COFs.[29 ] It is possible that these COFs lose their crystallinity under different mechanisms
(i.e., pore collapse or delamination/aggregation) depending on the type of monomer
used. This will be an area of further study by our group in the future.
Conclusions
In summary, we have synthesized three different hydrazone COFs (TFPB-DHz COF, TFPT-DHz COF, and Py-DHz COF) from unsubstituted hydrazide linkers. This study shows that having an alkoxy
group is not essential for the formation of 2D sheets or the initial stacking of those
sheets in hydrazone COFs. However, the crystallinity and porosity are significantly
affected by the choice of activation method. Furthermore, by introducing the potential
for greater flexibility into the COF backbone, additional functionality, such as dynamic
adsorption behavior, can be imagined for hydrazone-linked COFs. We found that gentler
scCO2 activation is an easy, efficient, and effective way to improve crystallinity and
porosity of hydrazone COFs and may open up access to more hydrazone COF structures
that were previously thought to be inaccessible.
Experimental Section
All the chemicals were purchased from commercially available sources (Fisher scientific,
TCI, Acros, Alfa Aesar) and they were used in reactions as received unless otherwise
mentioned. DHz was purchased from TCI. 5'-(4-Formylphenyl)-[1,1':3',1''-terphenyl]-4,4''-dicarbaldehyde
(TFPB ) and 4,4',4''-(1,3,5-triazine-2,4,6-triyl)tribenzaldehyde (TFPT ) were synthesized according to previously reported literature procedures.[18 ]
[21 ] The 1 H and 13 C NMR spectra for all the synthesized compounds were recorded on a Bruker Avance 600 MHz
spectrometer. The FT-IR spectra of monomers, model compound, and COFs were obtained
with a Cary 600 Series FT-IR spectrophotometer with an ATR attachment. The PXRD experiments
were carried out using a Bruker D8 Advance diffractometer with a sealed tube radiation
source (Cu Kα, λ = 1.54184 Å), a low background sample holder, and a Lynxeye XE detector.
Ultra-high purity grade N2 and CO2 gases were purchased from Airgas Corporation. A Micromeritics ASAP 2020 surface area
analyzer was used to perform low-pressure N2 adsorption–desorption (up to 760 torr) experiments of COFs. All analyses were carried
out at 77 K using a liquid N2 bath. To calculate the BET surface area of COFs, the data in the range of 0.01 < P /P
0
2 at 77 K and Cylindrical Pores in an Oxide Surface model in the Micromeritics software
package was used to determine the pore size distributions of COFs. scCO2 activation was performed using a Leica EM CPD 300 Critical Point Dryer.
Procedures
Synthesis of TFPB-DHz COF
TFPB (17 mg, 0.043 mmol) and DHz (12.7 mg, 0.065 mmol) were kept in a 5 mL ampoule along with mesitylene (0.9 mL)
and dioxane (0.1 mL). Then, the mixture was sonicated for 10 min followed by addition
of 6 M acetic acid (0.1 mL, aq.). After that, the mixture was flash-frozen in liquid
N2 and flame-sealed. Once the ampoule was warmed to room temperature, it was kept in
an oven at 120 °C for 72 h without any disturbances. Next, the ampoule was cooled
to room temperature and the precipitate was collected by filtration. The resulted
solid was washed with THF and ethanol. Then the wet filter cake was subjected to scCO2 activation to afford 20.4 mg (69%) of a colored powder of TFPB-DHz COF.
Synthesis of TFPT-DHz COF
TFPT (15 mg, 0.038 mmol) and DHz (11.1 mg, 0.057 mmol) were kept in a 5 mL ampoule along with mesitylene (0.9 mL)
and dioxane (0.1 mL). Then the mixture was sonicated for 10 min followed by addition
of 6 M acetic acid (0.1 mL, aq). After that, the mixture was flash-frozen in liquid
N2 and flame-sealed. Once the ampoule was warmed to room temperature, it was kept in
an oven at 120 °C for 72 h. Next, the ampoule was cooled to room temperature and the
precipitate was collected by filtration. The resulted solid was washed with THF and
ethanol. Then the wet filter cake was subjected to scCO2 activation to afford 14 mg (54%) of a yellow colored powder of TFPT-DHz COF.
Synthesis of Py-DHz COF
Py (16 mg, 0.026 mmol) and DHz (10.1 mg, 0.052 mmol) were kept in a 5 mL ampoule along with o -dichlorobenzene (0.9 mL) and n -butanol (0.1 mL). Then the mixture was sonicated for 10 min followed by addition
of 6 M acetic acid (0.1 mL, aq). After that, the mixture was flash-frozen in liquid
N2 and flame-sealed. Once the ampoule was warmed to room temperature, it was kept in
an oven at 120 °C for 120 h. Next, the ampoule was cooled to room temperature and
the precipitate was collected by filtration. The resulted solid was washed with THF
and ethanol. Then the wet filter cake was subjected to scCO2 activation to afford 13.3 mg (51%) of a yellow colored powder of Py-DHz COF.
Synthesis of Model Compound
DHz (250 mg, 1.29 mmol) and benzaldehyde (0.26 mL, 2.57 mmol) were placed in a round
bottom flask. Then, ethanol (10 mL) was added, and the mixture was refluxed for 24 h.
After that, the reaction mixture was allowed to cool to room temperature and the precipitate
was filtered off. Finally, the obtained solid was washed with ethanol and vacuum-dried
to afford pure product as white solid (406 mg, 85%). 1 H NMR (DMSO 600 MHz): δH 11.99 (s, 2 H), 8.49 (s, 2 H), 8.06 (s, 4 H), 7.76 (d, 4 H), 7.47 (m, 6 H) δC 162.88, 148.81, 136.59, 134.70, 130.71, 129.37, 128.27, 127.68
Supercritical Carbon Dioxide Drying of COFs
scCO2 activation was performed using a Leica EM CPD 300 Critical Point Dryer. After the
reaction time, the COF powders were filtered and washed thoroughly with THF and ethanol.
During filtration and washing, care was taken to avoid complete drying the COF powder.
The wet samples were placed into the scCO2 sample holders and were washed with scCO2 Program parameters: cooling temperature to keep CO2 fluid was 17 °C, the speed of CO2 influx in pressure chamber was set to slow, the exchange speed was set at 5, and
the number of cycles was set to 99. The heating speed for critical point was set to
medium, the temperature was set at 35 °C, and the gas release speed was set to medium.
